The first and most obvious requirement is that they would need light sensitivity, so vision is a must.
Beyond that there are some differences between the way light and sound work, for example hearing is 360 degree coverage, if someone shouts behind you then you can still see it. Equally sound can be used for communication around corners or over hills.
As a result light is actually quite weak to use as an early warning signal, so it is likely that there would be some auditory communication as well even if it's restricted to alarm and distress signals.
One advantage of visual signalling is the available bandwidth, if you had a creature with the ability to change its skin dramatically and rapidly in various patterns and shapes then it could well be able to transmit information far faster and at longer ranges than speech.
Multiple creatures could all be "speaking" in the same area at the same time without interference and conversations would not need people to wait for each person to finish. Instead even in a multi-way conversation each person could constantly talk all the time, so expect them to be better at multi-tasking and processing multiple information streams simultaneously than humans are. In general their visual processing would be superior to humans but their auditory processing would be weaker.
Technology would take different routes from our own, as the telephone and radio would both be pointless. Early remote communication would most likely be something like the telegraph and writing could well be hieroglyphics designed to look like the skin patterns used to "speak".
They may well develop independently mobile eyes so that they can watch people they are talking to at the same time as working. For example a conversation while sewing would require two eyes to watch the sewing process and another to watch the conversation. A group conversation may require either one eye on each person in the conversation or some sort of signal that causes everyone to switch attention to the person who is going to speak.